An intelligent method for cleaning alfalfa seeds
By using an intelligent alfalfa seed cleaning method that combines AI vision and near-infrared spectroscopy for multi-dimensional recognition and targeted pneumatic sorting, the problem of insufficient separation of near-characteristic impurities and mechanical damage in existing alfalfa seed cleaning technologies has been solved, achieving a high-purity, low-damage seed quality improvement.
Patent Information
- Application Number
- CN202511388301.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2025-09-26
- Publication Date
- 2026-11-10
- Estimated Expiration
- 2045-09-26
AI Technical Summary
Existing alfalfa seed cleaning methods have limited ability to separate impurities with similar characteristics, and mechanical force can easily cause physical damage to the seeds, leading to a decline in seed quality.
An intelligent alfalfa seed cleaning method is adopted, which utilizes AI vision and near-infrared spectroscopy to target pneumatic sorting. Through flexible screens and low-friction spiral conveying, combined with multi-layer screening and fine air separation, a seed quality correlation database is established to achieve adaptive parameter control.
It improved the separation accuracy of near-characteristic impurities to 99.5%, reduced the seed coat damage rate to below 1%, increased the germination rate by 10%-15%, increased the number of effective plants in the field by 20%, and increased the fresh grass yield by 15%-20%.
Smart Images

Figure CN120920353B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of alfalfa seed production technology, specifically to an intelligent alfalfa seed cleaning method. Background Technology
[0002] Alfalfa is known as the "king of forage" because it is not only high-yielding but also of excellent quality, making it a favorite food for various livestock and poultry. With the development of the commodity economy, the alfalfa industry has grown rapidly, and the planting area of alfalfa is expanding, especially in North China, where alfalfa has been widely planted. However, the current methods for cleaning alfalfa seeds still have the following technical problems: Existing cleaning methods mainly rely on the differences in physical properties between seeds and impurities, but their ability to separate impurities with similar properties is limited. During the cleaning process, mechanical force can easily cause physical damage to the seeds. Alfalfa seeds have thin seed coats, and if the cleaning equipment rotates too fast, the screen is too hard, or the gap is unreasonable, it can easily lead to seed coat rupture and endosperm damage. Excessive friction between the blades and seeds in some spiral conveyor cleaning equipment can damage the internal physiological structure of the seeds. Therefore, the shortcomings of these cleaning methods can seriously lead to a decline in the quality of alfalfa seeds.
[0003] Therefore, an intelligent alfalfa seed cleaning method is proposed to solve the problems mentioned above. Summary of the Invention
[0004] The purpose of this invention is to provide an intelligent alfalfa seed cleaning method to solve the problems mentioned in the background art.
[0005] To achieve the above objectives, the present invention provides the following technical solution: an intelligent alfalfa seed cleaning method, the specific steps of which are as follows: Step 1: Initial screening and pre-separation of impurities: The alfalfa seeds to be cleaned are evenly fed into the pretreatment unit through a vibrating cloth feeder. They are first coarsely screened through a 5mm aperture elastic screen to remove large particles of impurities such as straw and soil clods. The screen vibration frequency is set to 30-40Hz. After coarse screening, the seeds enter the negative pressure air separation chamber, where light impurities are separated by air force. A laser particle size sensor is installed in the air separation chamber to detect the difference in suspension velocity between impurities and seeds in the airflow in real time. If the residual rate of light impurities is >3%, the sensor feeds back the data and automatically adjusts the wind speed to 10-12m / s. Step 2: Establishment of Seed Characteristic Database: Take 100g of pretreated seed sample, scan with a near-infrared spectrometer to obtain the characteristic spectrum of alfalfa seeds, and simultaneously use a high-resolution industrial camera to photograph the seed morphology. Upload the data to the AI recognition system to establish a standard database of healthy alfalfa seeds; simultaneously record the spectral and morphological data of common near-characteristic impurities as a benchmark for identification and comparison. Step 3: Dynamic Identification of Near-Characteristic Impurities: The pre-treated seeds enter the intelligent detection channel via a belt conveyor. The channel integrates a dual-view industrial camera and a near-infrared spectral probe. Images and spectral data are simultaneously acquired as each seed passes through. The data is transmitted to the AI system in real time, and the deep learning model compares the seeds with healthy seeds and near-characteristic impurities in the database. Distinguish weed seeds of similar size; differentiate impurities with similar composition through near-infrared spectral differences; Step 4: Targeted Separation of Near-Characteristic Impurities: Seeds identified as near-characteristic impurities by AI are blown into a side impurity collection box by the system triggering a high-pressure airflow at the corresponding position when passing through the pneumatic sorting valve. Healthy seeds continue to the next stage with the conveyor belt. After sorting, the system passes through an online sampling inspection module. If the impurity residue rate is >0.5%, the system automatically optimizes the identification model parameters to ensure a separation accuracy of ≥99.5%. Step 5, Flexible Grading and Screening: The separated healthy seeds enter a multi-layer flexible screen unit. The screen apertures from top to bottom are 2.5mm, 2.0mm, and 1.0mm, respectively. The screen material is modified polyethylene, and the edges of the mesh are rounded. The screen vibration is driven by a servo motor, and the vibration frequency and amplitude are controlled in real time by a seed damage sensor. If the sensor detects a damage rate >1%, the frequency and amplitude are automatically reduced. Step 6, Low-friction screw conveyor: The screened seeds enter the flexible screw conveyor. The screw blades are coated with polytetrafluoroethylene, and the blade spacing is set to 3 times the seed length to avoid the seeds being squeezed. The conveying speed is monitored by a torque sensor. If the torque is >5 N·m, the speed is automatically reduced and the blade spraying device is activated. Step 7, Fine air separation and secondary purification: The transported seeds enter the fine air separation tower. The air speed is dynamically adjusted by the seed suspension velocity tester. The residual light impurities are separated by airflow. The inner wall of the air separation tower is lined with a sponge buffer layer. Step 8, Quality Inspection: Randomly select cleaned seed samples and test the germination rate using an automatic germination rate tester; observe the seed coat integrity using a microscope; re-check the purity using a near-infrared spectrometer; upload the test data to the system. If the germination rate is <90%, the damage rate is >2%, and the purity is <98%, the batch is automatically marked as unqualified, triggering parameter traceability analysis; the system optimizes the parameters of the preceding steps based on the quality inspection data. Establish a database linking cleaning parameters and seed quality, and use machine learning to predict the optimal parameter combination to achieve adaptive adjustment of parameters between batches; Step 9: Finished product collection and pretreatment: Qualified seeds are dried in a low-temperature air dryer to remove surface moisture, and then sent to a sterile collection chamber. Ultraviolet germicidal lamps are installed in the chamber to further reduce surface pathogens.
[0006] Preferably, in step one, the 5mm aperture elastic screen is made of polyurethane with a hardness of 60±5 Shore A; it removes large particles of straw and soil with a diameter > 5mm; and the initial wind speed of the negative pressure air separation chamber is set to 8-10m / s.
[0007] Preferably, in step two, the characteristic spectrum of alfalfa seeds is obtained by scanning with a near-infrared spectrometer, including characteristic peaks of protein and fat components; the morphology of the seeds, including size, color, and surface texture, is captured by a high-resolution industrial camera; and the alfalfa seed standard database contains morphological parameters of 1.5-2.5g per thousand seeds, 2.0-2.5mm in length, and 1.0-1.2mm in width, as well as characteristic spectral ranges.
[0008] Preferably, in step three, the conveyor belt of the belt conveyor line is made of food-grade silicone material with a surface friction of 0.3-0.4; the acquisition frequency of the integrated dual-view industrial camera is 500 frames / second.
[0009] Preferably, in step four, the pressure of the high-pressure airflow is 0.2-0.3 MPa, and the airflow duration is 0.1 s; the online sampling module automatically takes 100 samples every 5 minutes.
[0010] Preferably, in step five, the elastic modulus of the sieve is 200-300 MPa, and the seed damage sensor is installed below the sieve to detect the fluorescent substance released by the seed coat rupture; if the sensor detects a damage rate >1%, the frequency is automatically reduced from 40 Hz to 30 Hz; the amplitude is reduced from 5 mm to 3 mm.
[0011] Preferably, in step six, the friction coefficient of the spiral blade is 0.04-0.05; if the torque is >5 N·m, the conveying speed automatically decreases from 30 r / min to 20 r / min.
[0012] Preferably, in step seven, the thickness of the sponge buffer layer lining the inner wall of the air separation tower is 5mm.
[0013] Preferably, in step eight, when randomly selecting cleaned seed samples, three samples are taken from each batch, each weighing 100g; the constant temperature and humidity of the automatic germination rate tester are 25℃ and 80% during testing.
[0014] Preferably, in step nine, the temperature of the low-temperature air dryer is 35±2℃ and the wind speed is 2m / s; the surface moisture is removed by the low-temperature air dryer, reducing its moisture content to 8%-10%.
[0015] Compared with the prior art, the beneficial effects of the present invention are: 1) Separation of near-characteristic impurities: This application uses AI vision + near-infrared spectroscopy for multi-dimensional recognition, combined with targeted pneumatic sorting, to achieve a separation accuracy of over 99.5% for weed seeds and bacteria-carrying seeds that are similar in size and specific gravity, thereby improving the purity of alfalfa seeds; 2) Mechanical damage control: This application reduces the seed coat damage rate to below 1% by using a flexible screen, a low-friction spiral, and adaptive parameter adjustment, while keeping the endosperm and internal structure intact, increasing the germination rate by 10%-15%, resulting in vigorous seedling growth and enhanced stress resistance. 3) Increased yield and quality: This application increases the number of effective plants in the field by more than 20% and reduces the weed rate by 30% after sowing high-purity (≥98%) and low-damage seeds. Furthermore, due to the high seed activity, alfalfa branches more and fresh grass yield increases by 15%-20%. At the same time, it reduces the quality decline caused by diseased seeds. Attached Figure Description
[0016] Figure 1 This is a flowchart of the steps in this method. Detailed Implementation
[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0018] Example: Please see Figure 1 The present invention provides a technical solution: An intelligent alfalfa seed cleaning method, the specific steps of which are as follows: Step 1: Initial screening and pre-separation of impurities: The alfalfa seeds to be cleaned are evenly fed into the pretreatment unit by a vibrating cloth feeder. They are first coarsely screened through a 5mm aperture elastic screen to remove large particles of impurities such as straw and soil clods. The screen vibration frequency is set to 30-40Hz and monitored in real time by a vibration sensor to avoid excessive seed bouncing due to resonance. After coarse screening, the seeds enter the negative pressure air separation chamber, where light impurities are separated by air. A laser particle size sensor is installed in the air separation chamber to detect the difference in suspension velocity between impurities and seeds in the airflow in real time. If the residual rate of light impurities is >3%, the sensor feeds back the data and automatically adjusts the wind speed to 10-12m / s. Step 2: Establishment of Seed Characteristic Database: Take 100g of pretreated seed sample, scan with a near-infrared spectrometer to obtain the characteristic spectrum of alfalfa seeds, and simultaneously use a high-resolution industrial camera to photograph the seed morphology. Upload the data to the AI recognition system to establish a standard database of healthy alfalfa seeds; simultaneously record the spectral and morphological data of common near-characteristic impurities as a benchmark for identification and comparison. Step 3: Dynamic Identification of Near-Characteristic Impurities: The pre-treated seeds enter the intelligent detection channel via a belt conveyor. The channel integrates a dual-view industrial camera and a near-infrared spectral probe. Images and spectral data are simultaneously acquired as each seed passes through. The data is transmitted to the AI system in real time, and the deep learning model compares the seeds with healthy seeds and near-characteristic impurities in the database. Distinguish weed seeds of similar size; differentiate impurities with similar composition through near-infrared spectral differences; Step 4: Targeted Separation of Near-Characteristic Impurities: Seeds identified as near-characteristic impurities by AI are blown into a side impurity collection box by the system triggering a high-pressure airflow at the corresponding position when passing through the pneumatic sorting valve. Healthy seeds continue to the next stage with the conveyor belt. After sorting, the system passes through an online sampling inspection module. If the impurity residue rate is >0.5%, the system automatically optimizes the identification model parameters to ensure a separation accuracy of ≥99.5%. Step 5, Flexible Grading and Screening: The separated healthy seeds enter a multi-layer flexible screen unit. The screen apertures from top to bottom are 2.5mm (removing slightly larger impurities), 2.0mm (retaining qualified seeds), and 1.0mm (collecting fine particles). The screen material is modified polyethylene (elastic modulus 200-300MPa), and the edges of the mesh are rounded with a radius of 0.1mm to avoid scratching the seed coat. The screen vibration is driven by a servo motor, and the vibration frequency and amplitude are controlled in real time by a seed damage sensor: if the sensor detects a damage rate >1%, the frequency and amplitude are automatically reduced. Step 6, Low-friction spiral conveyor: The screened seeds enter the flexible spiral conveyor. The spiral blades are coated with polytetrafluoroethylene, and the blade spacing is set to 3 times the seed length to avoid the seeds being squeezed. The conveying speed is monitored by a torque sensor. If the torque is >5 N·m, it indicates that the seed accumulation friction is increasing, the speed is automatically reduced, and the blade spraying device is activated. Step 7, Fine air separation and secondary purification: The transported seeds enter the fine air separation tower. The air speed is dynamically adjusted by the seed suspension velocity tester. The residual light impurities are separated by airflow. The inner wall of the air separation tower is lined with a sponge buffer layer. Step 8, Quality Inspection: Randomly select cleaned seed samples and test the germination rate using an automatic germination rate tester; observe the seed coat integrity using a microscope; re-check the purity using a near-infrared spectrometer; upload the test data to the system. If the germination rate is <90%, the damage rate is >2%, and the purity is <98%, the batch is automatically marked as unqualified, triggering parameter traceability analysis; based on the quality inspection data, the system reverse-optimizes the parameters of the preceding steps: for example, if the seed coat damage rate is high, adjust the screen hardness (increase by 5 Shore A) and reduce the spiral speed; if there are many near-characteristic impurities remaining, update the spectral comparison threshold of the AI recognition model.
[0019] Establish a database linking cleaning parameters and seed quality, and use machine learning to predict the optimal parameter combination. For example, for seeds with a thousand-seed weight of 2.0g, it is recommended to have a screen frequency of 35Hz, a screw speed of 25r / min, and a wind speed of 7m / s, so as to achieve adaptive adjustment of parameters between batches. Step 9: Finished product collection and pretreatment: Qualified seeds are dried in a low-temperature air dryer to remove surface moisture, and then sent to a sterile collection chamber. Ultraviolet germicidal lamps are installed in the chamber to further reduce surface pathogens.
[0020] In step one, the selected 5mm aperture elastic screen is made of polyurethane with a hardness of 60±5 Shore A; it removes large particles of straw and soil with a diameter >5mm; the initial wind speed of the negative pressure air separator is set to 8-10m / s; its function is as follows: The 5mm aperture elastic screen is made of polyurethane with a hardness of 60±5 Shore A. The purpose of this material is that polyurethane is wear-resistant, impact-resistant, and highly elastic. With a hardness of 60±5 Shore A, the screen can maintain a certain level of rigidity while possessing good elastic deformation capability. In the coarse screening of alfalfa seeds, this material and hardness effectively withstand the impact of falling seeds and impurities, reducing screen wear and extending its service life. It also allows seeds to pass through the screen openings more easily, improving screening efficiency and avoiding poor screening results or seed damage caused by screens that are too hard or too soft. The purpose of setting the screen vibration frequency to 30-40Hz: Vibration frequency is a crucial factor affecting screening efficiency. At low frequencies, particles cannot be effectively thrown up, easily clogging the screen openings and resulting in low screening efficiency. Conversely, at high frequencies, the particles move too actively on the screen, increasing collisions between particles and between particles and the screen, potentially preventing particles from falling smoothly and also impacting screening efficiency. Setting the screen vibration frequency to 30-40Hz allows alfalfa seeds to maintain good agitation on the screen surface, facilitating the separation of large impurities from the seeds, reducing seed residence time on the screen, increasing screening speed, and preventing damage to the seeds from excessively high vibration frequencies.
[0021] In step two, the characteristic spectrum of alfalfa seeds is obtained by scanning with a near-infrared spectrometer, including characteristic peaks of protein and fat components; the morphology of the seeds, including size, color, and surface texture, is captured by a high-resolution industrial camera; the alfalfa seed standard database contains morphological parameters of 1.5-2.5g per thousand seeds, 2.0-2.5mm in length, and 1.0-1.2mm in width, as well as characteristic spectral ranges.
[0022] In step three, the conveyor belt of the belt conveyor line is made of food-grade silicone material with a surface friction of 0.3-0.4; the integrated dual-view industrial camera has a sampling frequency of 500 frames / second; its function is as follows: The purpose of using food-grade silicone in belt conveyor lines: The selection of food-grade silicone is key to balancing seed safety, conveying stability, and equipment durability. Specific functions include: Ensuring seeds are free from contamination and damage: The core characteristics of food-grade silicone are that it is non-toxic and free of chemical residues, preventing the release of harmful substances from the conveyor belt material that could contaminate alfalfa seeds and ensuring the safety of subsequent germination and planting processes; at the same time, silicone material has good flexibility, which can significantly reduce the collision and wear of seeds during transportation compared with rigid conveyor belts, reduce the risk of seed coat damage, and indirectly ensure the germination rate. Adapt to cleaning environment and reduce maintenance costs: Dust is generated during the cleaning process and there may be slight humidity fluctuations. Food-grade silicone has the characteristics of wear resistance, dust adhesion resistance, temperature resistance and slight corrosion resistance. It is not easy to cause conveyor jamming due to dust accumulation, nor is it easy to age and crack due to changes in environmental humidity. It can extend the service life of the conveyor belt and reduce the frequency and cost of maintenance in industrial continuous production. The function of the conveyor belt surface friction force of 0.3-0.4: This friction force range is a dynamically balanced value verified by experiments. It avoids seed slippage that affects detection accuracy, while also preventing excessive friction from damaging the seeds or hindering subsequent sorting. Specific functions include: To prevent seeds from slipping and ensure stable detection position: If the friction force is less than 0.3, the surface of the conveyor belt is too smooth, and the seeds are prone to slipping and displacement during continuous conveying, causing the seed position to deviate from the detection area when the dual-view industrial camera captures images; AI recognition relies on accurate feature extraction of seed morphology, and positional deviation will directly lead to feature comparison deviation, making it impossible to accurately distinguish weed seeds of similar size. To avoid excessive friction and ensure easy seed detachment and integrity: If the friction force is >0.4, the adhesion between the seed and the conveyor belt surface will be too strong, which will cause two problems: the seed will have difficulty smoothly entering the subsequent pneumatic sorting valve area from the conveyor belt, and may be stuck on the conveyor belt, resulting in missed sorting; the increased friction between the seed and the conveyor belt will easily wear down the seed coat, resulting in the seed coat integrity test in step eight failing to meet the standard, or affecting the germination rate. Adaptable to conveyor speed, enabling stable batch testing: The friction force of 0.3-0.4 can match the normal operating speed of belt conveyor lines, ensuring that each seed can pass through the intelligent testing channel at a uniform speed without displacement, providing a stable foundation for the synchronous data acquisition of dual-view cameras and near-infrared spectral probes.
[0023] The role of a dual-view industrial camera with a 500 frames per second acquisition frequency: High-frequency acquisition of 500 frames per second is a key design feature for accurate imaging of dynamic seeds, directly determining the accuracy of AI in identifying near-characteristic impurities. Specific functions include: Eliminating motion blur and capturing clear morphological details of seeds: Alfalfa seeds are extremely small and move continuously with the conveyor belt. If the acquisition frequency is too low, insufficient shutter speed will cause image blurring, making it impossible to clearly capture the surface texture and color differences of the seeds. These details are the core basis for AI to distinguish impurities with similar characteristics, and blurred images will directly lead to misjudgment. Matching dual-view synchronous acquisition ensures complete multi-dimensional features: The dual-view camera is designed to acquire three-dimensional morphological data of the seed's front and side views, while the high-frequency acquisition of 500 frames per second ensures that the images from the two views are completely synchronized in time, avoiding viewpoint misalignment due to insufficient frame rate, and ensuring that AI can compare the multi-dimensional morphological data of the same seed with the database to improve recognition accuracy.
[0024] In step four, the pressure of the high-pressure airflow is 0.2-0.3 MPa, and the airflow duration is 0.1 s; the online sampling module automatically takes 100 samples every 5 minutes.
[0025] In step five, the elastic modulus of the sieve is 200-300MPa, and the seed damage sensor is installed below the sieve to detect the fluorescent substances released by the seed coat rupture. If the sensor detects a damage rate >1%, the frequency is automatically reduced from 40Hz to 30Hz, and the amplitude is reduced from 5mm to 3mm.
[0026] In step six, the friction coefficient of the helical blades is 0.04-0.05; if the torque is greater than 5 N·m, the conveying speed automatically decreases from 30 r / min to 20 r / min; its function is as follows: The function of a friction coefficient of 0.04-0.05 for spiral blades is to achieve low-friction transport and protect the integrity of seeds.
[0027] The core risk of spiral conveying is seed coat damage caused by friction between the leaves and seeds. This friction coefficient range is an experimentally verified critical value for low damage. Its specific function is as follows: Minimize seed coat wear to ensure the foundation for seed germination: If the coefficient of friction is >0.05, the friction between the leaf and the seed surface is too great, which will lead to two problems: when the seed rotates with the leaf, the seed coat is repeatedly rubbed and scraped, resulting in scratches or damage, and the seed coat integrity test is likely to fail; the damaged seed coat may cause the seed to absorb moisture and become moldy, reducing the germination rate test pass rate.
[0028] A low coefficient of friction of 0.04-0.05 can significantly reduce frictional damage between leaves and seeds, ensuring that the seed coat integrity meets the standards and providing a basic guarantee for subsequent germination rate; To prevent seed sticking and ensure smooth, unobstructed transport: Alfalfa seeds are small and may contain trace amounts of moisture. If the coefficient of friction is too high, the seeds easily stick to the leaf surface, causing blockages in the transport channel and affecting batch cleaning efficiency. Sticky seeds are further damaged by subsequent seeds being squeezed. A low coefficient of friction allows the seeds to quickly detach from the leaves as they rotate, moving smoothly along the transport channel, avoiding sticking or blockages, and ensuring the stability of continuous industrial production. Reducing energy consumption and blade wear, and decreasing maintenance costs, is crucial. The coefficient of friction is directly related to transport energy consumption; a higher coefficient of friction means greater resistance that the motor must overcome to drive the blades, resulting in higher energy consumption. Simultaneously, high friction accelerates blade wear and shortens blade lifespan. A friction coefficient of 0.04-0.05 can reduce motor load by approximately 30%-40%, reducing energy consumption while extending the lifespan of the PTFE coating and lowering long-term maintenance costs.
[0029] The purpose of reducing the speed from 30 r / min to 20 r / min when the torque is >5 N·m is to achieve overload protection and balance conveying efficiency and safety. Torque is a core indicator reflecting the load size of the screw conveyor. The design of reducing the speed is essentially to dynamically adapt to the load and avoid double damage to the equipment and the conveyor. Specific functions include: To prevent motor overload damage and ensure safe equipment operation: The motor power of the screw conveyor should be matched with the rated torque. If the torque is >5 N·m and the speed is maintained at 30 r / min, the motor needs to output more power to overcome the resistance. Long-term operation will lead to motor overheating and burnout, or gearbox wear. After the speed is reduced to 20 r / min, the motor speed decreases and the output torque demand decreases accordingly, avoiding motor overload and ensuring stable equipment operation.
[0030] Reduce seed compression damage and avoid mass scrapping: The core reason for torque > 5 N·m is excessive seed accumulation in the conveying channel. If a high speed is maintained at this time, the accumulated seeds will be squeezed by both the blades and the conveying pipe wall, resulting in seed deformation; broken seed powder mixes with whole seeds, contaminating the finished product and reducing the purity test pass rate. When the rotation speed is reduced to 20 r / min, the blade rotation speed slows down, which allows time for the accumulated seeds to disperse, reduces the probability of crushing, controls the seed breakage rate to below 1%, and avoids batches of seeds being scrapped due to crushing.
[0031] In step seven, the inner wall of the air separator is lined with a 5mm thick sponge buffer layer to prevent seed impact damage.
[0032] In step eight, when randomly selecting cleaned seed samples, three samples are taken from each batch, each weighing 100g; the automatic germination rate tester is used at a constant temperature of 25℃ and a humidity of 80%.
[0033] In step nine, the temperature of the low-temperature air dryer is 35±2℃ and the wind speed is 2m / s; the surface moisture is removed by the low-temperature air dryer, reducing its moisture content to 8%-10%.
[0034] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or basic characteristics. Therefore, the embodiments should be considered exemplary and non-limiting in all respects. The scope of the invention is defined by the appended claims rather than the foregoing description. Therefore, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention, and no reference numerals in the claims should be construed as limiting the scope of the claims.
[0035] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.
Claims
1. An intelligent method for cleaning alfalfa seeds, characterized in that, The specific steps of this intelligent alfalfa seed cleaning method are as follows: Step 1: Initial screening and pre-separation of impurities: The alfalfa seeds to be cleaned are evenly fed into the pretreatment unit through a vibrating cloth feeder. They are first coarsely screened through a 5mm aperture elastic screen to remove large particles of impurities such as straw and soil clods. The screen vibration frequency is set to 30-40Hz. After coarse screening, the seeds enter the negative pressure air separation chamber, where light impurities are separated by air force. A laser particle size sensor is installed in the air separation chamber to detect the difference in suspension velocity between impurities and seeds in the airflow in real time. If the residual rate of light impurities is >3%, the sensor feeds back the data and automatically adjusts the wind speed to 10-12m / s. Step 2: Establishment of Seed Characteristic Database: Take 100g of pretreated seed sample, scan with a near-infrared spectrometer to obtain the characteristic spectrum of alfalfa seeds, and simultaneously use a high-resolution industrial camera to photograph the seed morphology. Upload the data to the AI recognition system to establish a standard database of healthy alfalfa seeds; simultaneously record the spectral and morphological data of common near-characteristic impurities as a benchmark for identification and comparison. Step 3: Dynamic Identification of Near-Characteristic Impurities: The pre-treated seeds enter the intelligent detection channel via a belt conveyor. The channel integrates a dual-view industrial camera and a near-infrared spectral probe. Images and spectral data are simultaneously acquired as each seed passes through. The data is transmitted to the AI system in real time, and the deep learning model compares the seeds with healthy seeds and near-characteristic impurities in the database. Distinguish weed seeds of similar size; differentiate impurities with similar composition through near-infrared spectral differences; Step 4: Targeted Separation of Near-Characteristic Impurities: Seeds identified as near-characteristic impurities by AI are blown into a side impurity collection box by the system triggering a high-pressure airflow at the corresponding position when passing through the pneumatic sorting valve. Healthy seeds continue to the next stage with the conveyor belt. After sorting, the system passes through an online sampling inspection module. If the impurity residue rate is >0.5%, the system automatically optimizes the identification model parameters to ensure a separation accuracy of ≥99.5%. Step 5, Flexible Grading and Screening: The separated healthy seeds enter a multi-layer flexible screen unit. The screen apertures from top to bottom are 2.5mm, 2.0mm, and 1.0mm, respectively. The screen material is modified polyethylene, and the edges of the mesh are rounded. The screen vibration is driven by a servo motor, and the vibration frequency and amplitude are controlled in real time by a seed damage sensor. If the sensor detects a damage rate >1%, the frequency and amplitude are automatically reduced. Step 6, Low-friction screw conveyor: The screened seeds enter the flexible screw conveyor. The screw blades are coated with polytetrafluoroethylene, and the blade spacing is set to 3 times the seed length to avoid the seeds being squeezed. The conveying speed is monitored by a torque sensor. If the torque is >5 N·m, the speed is automatically reduced and the blade spraying device is activated. Step 7, Fine air separation and secondary purification: The transported seeds enter the fine air separation tower. The air speed is dynamically adjusted by the seed suspension velocity tester. The residual light impurities are separated by airflow. The inner wall of the air separation tower is lined with a sponge buffer layer. Step 8, Quality Inspection: Randomly select cleaned seed samples and test the germination rate using an automatic germination rate tester; observe the seed coat integrity using a microscope; re-check the purity using a near-infrared spectrometer; upload the test data to the system. If the germination rate is <90%, the damage rate is >2%, and the purity is <98%, the batch is automatically marked as unqualified, triggering parameter traceability analysis; the system optimizes the parameters of the preceding steps based on the quality inspection data. Establish a database linking cleaning parameters and seed quality, and use machine learning to predict the optimal parameter combination to achieve adaptive adjustment of parameters between batches; Step 9: Finished product collection and pretreatment: Qualified seeds are dried in a low-temperature air dryer to remove surface moisture, and then sent to a sterile collection chamber. Ultraviolet germicidal lamps are installed in the chamber to further reduce surface pathogens.
2. The intelligent alfalfa seed cleaning method according to claim 1, characterized in that: In step one, the selected 5mm aperture elastic screen is made of polyurethane with a hardness of 60±5 Shore A; it removes large particles of straw and soil with a diameter > 5mm; the initial wind speed of the negative pressure air separation chamber is set to 8-10m / s.
3. The intelligent alfalfa seed cleaning method according to claim 1, characterized in that: In step two, the characteristic spectrum of alfalfa seeds is obtained by scanning with a near-infrared spectrometer, including characteristic peaks of protein and fat components; the morphology of the seeds, including size, color, and surface texture, is captured by a high-resolution industrial camera; the alfalfa seed standard database contains morphological parameters of 1.5-2.5g per thousand seeds, 2.0-2.5mm in length, and 1.0-1.2mm in width, as well as characteristic spectral ranges.
4. The intelligent alfalfa seed cleaning method according to claim 1, characterized in that: In step three, the conveyor belt of the belt conveyor line is made of food-grade silicone material with a surface friction of 0.3-0.4; the acquisition frequency of the integrated dual-view industrial camera is 500 frames / second.
5. The intelligent alfalfa seed cleaning method according to claim 1, characterized in that: In step four, the pressure of the high-pressure airflow is 0.2-0.3 MPa, and the airflow duration is 0.1 s; the online sampling module automatically takes 100 samples every 5 minutes.
6. The intelligent alfalfa seed cleaning method according to claim 1, characterized in that: In step five, the elastic modulus of the sieve is 200-300MPa, and the seed damage sensor is installed below the sieve to detect the fluorescent substances released by the seed coat rupture. If the sensor detects a damage rate >1%, the frequency is automatically reduced from 40Hz to 30Hz, and the amplitude is reduced from 5mm to 3mm.
7. The intelligent alfalfa seed cleaning method according to claim 1, characterized in that: In step six, the friction coefficient of the spiral blade is 0.04-0.05; if the torque is greater than 5 N·m, the conveying speed will automatically decrease from 30 r / min to 20 r / min.
8. The intelligent alfalfa seed cleaning method according to claim 1, characterized in that: In step seven, the thickness of the sponge buffer layer lining the inner wall of the air separation tower is 5mm.
9. The intelligent alfalfa seed cleaning method according to claim 1, characterized in that: In step eight, when randomly selecting cleaned seed samples, three samples are taken from each batch, each weighing 100g; the automatic germination rate tester is used at a constant temperature of 25℃ and a humidity of 80%.
10. The intelligent alfalfa seed cleaning method according to claim 1, characterized in that: In step nine, the temperature of the low-temperature air dryer is 35±2℃ and the wind speed is 2m / s; the surface moisture is removed by the low-temperature air dryer, reducing its moisture content to 8%-10%.
Citation Information
Patent Citations
A method for improving the crop seed germination rate by applying a deep learning convolutional neural network
CN109948531A
Seed sorting system
CN112024421A
